Contacter l'auteur / Contact the author

Recherche dans ce site / Search in this site







Software review

HFProp 1.3 main window.

HFProp propagation analysis and prediction program

Foreword. This review was written in 2004 for versions 1.2 and 1.3. Today these versions have been replaced by VOAProp that uses the VOACAP engine. It shows the same look and feel, and you can download it free of charge on G4ILO's website.

Developed by Julian Moss, G4ILO, HFProp is a quite simple propagation program that gives you quickly propagation estimations.

Before 1995 there was several free propagation prediction programs written in GWBasic, such as MiniMUF and MaxiMUF. However, the only free programs available for Windows were not very easy to use and took a lot of effort to produce a prediction for one particular path or circuit (point-to-point method).

In the end, Julian Moss wrote his own Windows application based on an old GWBasic code that he developed earlier. Then he began tweaking the algorithms to make them better approximate the results of other more complex programs.

As he told himself, he was never completely satisfied with the results and after a while he gave up the idea to optimize his application.

Recently, he rediscovered his program and started to play with it again. He tidied up a few things, tweaked the algorithms a bit more, and added a few new features to eventually release in 2003 a version 1.2 and a faster version 1.3 in 2004.

HFProp does not use any IONCAP, VOACAP or down-sized IRI model of the ionosphere; it is much simpler than this. It works essentially with an algorithm based on the F-layer model developed by Raymond Fricker of the BBC, a well-known algorithm already used in DOS programs like Miniprop or DXaid in the late '90s. It is completed with a geomagnetic model that predicts effects of the K-index at high latitudes.

Your propagation estimation in a few seconds

Positive side, highlight several points. First only a few seconds are necessary to generate a prediction map. Then, contrarily to other applications, HFProp asks explicitely for A- and K-index values and provides an automatic SFI to SSN converter. Unlike other programs, it shows no error in drawing short and long paths. Even very close to the frame and at 90 from your home location (e.g. near ZL at 39S and 179W by the long path) the path follows strictly geodesics and doesn't show erratic trajectories, steep angle and steps like some of its competitors.

HFProp 1.2 uses a nice cartesian world map highlighting cities plunged into darkness, on which is displayed the gray line and all accessible countries for a specified band, time and day. In addition I request to show the short path to Taiwan (BV) and get the bearing and his signal strength to the receiver. Clicking on the map, a MUF/LUF window poped up to display additional information.

That said, HFProp is much simpler to setup and to use that a program using a VOACAP engine. And I bet that most of you are not ready to spend minutes behind their screen to set a prediction that eventually will maybe not reflect your real working conditions. In this case HFProp is made or you, or almost !

HFProp deserves a try for several reasons. First it is free. Then contrarily to many other propagation prediction programs, it displays the MUF, LUF and critical frequencies (E and F2) using iso-contour maps incrusted in a gray line cartesian map of the world instead or line, bar graphs or text reports. In this way you see for example immediately the "ionization islands" along the equatorial anomaly and, correlated with the prediction for a specific band, you can estimate your chance to work such or such country, although the reliability is in the average.

In addition, in each band, in "DXer mode" in which the accuracy is best (this is a cursor to move) HFProp displays all entities (prefixes) accessible under the specified conditions. The intensity of each prefix varies according to the signal strength (white to gray). In addition, clicking on a location, an indicator located below right displays the signal strength in form of green and red LEDs. This map is self-explanatory and easy to use because there is no need to clic on the image to get a propagation chart like in other programs. If you do it, a new window pops up showing together the MUF and LUF variation to that remote station as displayed at left. However this map does not take into account the reliability or the antenna receive properties.

HFProp also includes a beacon tracker (offline simulation) as accurate as the one of its competitors that displays also their estimated signal strength.

At last in the new version 1.3, G4ILO added a link to the Internet. You can type any URL and get online data like WWV messages. In addition it included an automatic updater to get current solar flux and K index. In the new version 1.3, the two critical frequency charts for E and F2 layers are not displayed by default, but can be enabled (as can maps for other bands, including broadcast bands) by right-clicking the map and selecting "Set visible tabs". The version 1.3 is also became a "donationware"; as long as you don't donate some money to encourage Julian to develop his program, an "unregistered" message will be displayed on top of the screen.

Accuracy of forecasts

What can we expect from HFProp forecasts ? How does it perform against its competitors ? All depends on your needs. Programs using the VOACAP model to name the reference program are surely more more accurate and complete than HFProp but they also request more effort and are often difficult to master at first sight due to the definition of all parameters of a communication circuit. If HPprop is able to display a forecast in 5 seconds, VOACAP requests one minute or more if you master the program.

On his side, Julian is not sure that propagation forecasts can ever be very accurate, however complex the model : "trying to predict ionospheric conditions based on just two variables which were measured a few hours earlier is like trying to forecast the weather given just the temperature and humidity. Therefore I'm not convinced that lengthy, highly accurate calculations are necessarily worth the time they take".

It is a fact that HFProp uses very few variables compared to the ones included in a VOACAP model : 6 vs. 30.... HFProp ignores for example specifications of both transmitter and receiver terminals (antenna gain, power, takeoff angle, etc), it doesn't check ground properties (the good influence of the sea for example) or the change in propagation along the gray line. At last it doesn't take into account the noise level at receive, hop structure, S/I, S/N reliability, required reliability or the multipath tolerance to name most important parameters available in the VOACAP model. By design, its predictions cannot thus show the same accuracy as a VOACAP-based application.

Its results are displayed at the global earth scale with a geographical accuracy close to 5. Only the MUF/LUF chart can be predicted for a specified circuit, between your home and a target location.

HFProp works with statistical values, offline, and displays thus some difference over other programs when calculating the propagation to a DX station, or it does rough assumptions. In a sense it is an advantage because it gives you quickly an idea of the best times and frequencies for DXing what could be the other influences. But this method has also some drawbacks.

Most of the time if you work a DX station placed in the middle of the coverage, where usually the reliability is the highest, but not always, you will get the result that you expect (if at least amateurs are QRV in that remote country, Hi!). But consider the prediction displayed on the iso-contour frequency map or call signs incrusted as representing a rought estimation, a global view of the expected propagation conditions. Or said in other words, data displayed are not entirely reliable because the system is not able to calculate this reliability; it has no input data and no function to make this forecast or it assumes the highest value, what is not often realistic.

Take an example. Working on the air I am sure that you have already been the witness of a sudden 20-meter band closing down in the afternoon although it was wide open a few minutes earlier or a 40-meter band with deep QSB and noise during the morning. Both events are usually correlated with either secundary effects of a large sunflare with an increasing of planetary indices, a high LUF at daytime or even the darkness when you work close to the gray line.  How HFProp handles these effects ?

HFProp, like many similar programs, has limited prediction means. On the 40-m band for example at 0700 UTC at summertime, from a home location in ON, it displays a propagation chart extending over all western Europe including Newfoundland (VO1) with a good signal (at left). In fact in the field the propagation extended to Brazil also. The LUF was predicted to increase in frequency and at that time should exceed 7 MHz to reach 10 MHz at noon. Three hours later the band was predicted so closed that only the "DL" prefix was displayed (at right). In the field the propagation weakening as forecasted, and at short distance there was strong fading and noise during hours due to the LUF placed quite high. However, at 1000 UTC (right map) or even at noon, England, South France, Italy and even Yugoslavia that was placed where the LUF was predicted at 10 MHz where always reachable, but with noise. So the left prediction was also valid three to five hours later contrarily to its calculations. Worst, as usual, used without additional charts, the single MUF/LUFchart gave misleaded information.

Working offline with rather simple models, disturbances cannot be forecasted by HFProp, all the less when default values are placed much too high. I tried for example to know what reliability factor used the LUF algorithm that seemed to give a too high LUF forecast. I used the ICEPAC then the VOACAP model and a complete circuit between Brussels and Paris, path along which we experimented a strong QSB over 6 S-points[1] and a LUF at 10 MHz by noon. It appeared that the reliability was set to a value that even broadcasts do not use : to get a LUF as high I had to set both Circuit reliability and S/N reliability over 95% where amateurs usually set a SNR of 50% or even less in CW mode. Even with this value I never exceed a LUF of 8.7 MHz, and even so the reliability was not met. In fact the real SNR at noon was 38%, thus fair signals, and the LUF was never higher than 4 MHz. The high reliability used by default by HFProp affected thus the LUF estimation and all the propagation forecast. 

More interesting, the VOACAP model gave a signal power lower decile (taking into account the signal strength and loss) on the 40-m band decreasing gradually up to 6 dB around 0900 UTC, just when I worked these stations. Working with statistical values, these charts and figures must be interpreted with care and well cross-checked with other parameters. But in all cases this power loss of 75% (6 dB) represents a receive power dropping from 100W to less than... 25W. This is less than the the signal loss that I experimented, but it matches with the time at which all amateurs and I experimented a strong QSB on bands (of short period not shown in the chart) extending over tens of minutes that was impossible to forecast in HFProp. The same event, but inverted, occured by 2000 UTC and was correlated with a signal increase. But don't worry, if VOACAP predicted these fluctuations, HFProp is in the average; many other programs, even more expensive or sophisticated, are no more able to foresee these disturbances.

If HFprop doesn't work with many ionospheric models, it used in spite of everything Fricker's F-layer model and a geomagnetic model to take into account disturbed conditions. If HFProp is unable to foresee and to show you explicitely a short opening, fading, a disturbance (SID) or a polar absorption (PCA), it is however able to tell you that such or such area is unreachable due to the geomagnetic storm. In fact it doesn't tell you that explicitely but it removes all entity prefixes over the concerned area when K exceeds 5 or so.

In the field you will see that HFProp predictions are often too tight and pessimistic. But it is clear that using a handful variables injected in an basic ionospheric model, such a program cannot gives you very accurate predictions, all the less using offline statistical data, interpolations and knowing that ionospheric layers evolve, vanishing or consolidating in some tens of minutes, sometimes faster a the lowest level. HFProp is able to "find" the LUF increasing due to the sun presence for example but never the QSB lost in its inaccuracies.

From my trials and comparisons made with other propagation programs using or not a VOACAP engine, the error can exceed 6 hours in the time estimation, about 2 MHz in frequency and about 5000 km for the propagation chart (grayline map with incrusted call signs). Taken independently or even all together, these errors can easily lead to miss a DX station because the band closed earlier than expected or because the opening will only occur a few hours later, without speaking of the other parameters simply ignored by the system.

My final impression

Interface of VOAProp remplacing HFProp since 2004.

Some authors stated that forecasts provided by HFProp come close to predictions calculated by the IONCAP model. Personally, I have doubts about its ability to get the same accuracy as a program using a more complete ionospheric model and numerous parameters. Don't forget either that IONCAP (today replaced by VOACAP and other tools) is able to provide long-term availability figures, SNR and other reliability, as many output parameters unknown from HFProp. Hard to compare apples and pears in such conditions. The difference between forecasts established by the first over the second are thus obvious, especially when you compare their respective estimations for beacons, easily comparable between programs as they are always on the air at the same location and frequency.

For short, don't trust in HFProp forecasts with more than 50% of confidence, even though. Like many other programs, using very few algorithms, HFProp only make rough estimations using a very simple ionospheric model and an unrealistic reliability. Consequently, at one occasion the DX country displayed will be unreadable, at another occasion you will reach or hear a country 5000 km further that predicted. This result is thus still worse than having a 50-50 chance to work or not a DX station.

So, all well considered, HFProp has minuses and pluses. Globally, thanks to its nice world map based on satellite images and a user-friendly interface, the product is attracting at first look. Thanks to its original approach using iso-contour maps and incrusted data, HFProp can give you "the big picture" but it is soiled with a serious estimation error. Using a too simple model, it is unable to estimate whether a QSO is possible or not under the specified conditions, what at the end is the essential.

Now, if you only want to know in what direction to bear your antenna or if a "summary" map is all you request, go with it.

For more information

HFProp is free but no more available (but you can download version 1.3 from this site). Since 2004, this product has been replaced by VOAProp v1.1 that uses the VOACAP engine, and than you can download from G4ILO's website. The last corrections were made in 2008. The last version (2015) of the VOACAP engine can be downloaded from Greg Hand's website.

Back to Menu

[1] If you remember your studies of radioelectricity, 1 S-point represents a power ratio of 2 (e.g. 100 W/1000 W) or a 10 dB change (3 dB each time you double your power). In the same way a signal strength dropping of 6 dB is equivalent to a power loss of 75% (2x 3 dB or 2x 50% less in this case). 6 dB is also a power ratio of 4 times knowing that 3 dB = 2x, 6 dB = 4x, 9 dB = 8x, 10 dB = 10x, 20 dB = 100x.

Back to:


Copyright & FAQ